From f29896fc91a0f0fe8a3299553e38de58a76ad329 Mon Sep 17 00:00:00 2001
From: RichardG867 <richardg867@gmail.com>
Date: Sun, 10 Apr 2022 20:21:35 -0300
Subject: [PATCH] bios_extract: Keep partially-decompressed AMI modules around

---
 bios_extract/src/ami.c | 46 ++++++++++++++++++++++++++++++++++++------
 biostools/analyzers.py |  7 ++++---
 2 files changed, 44 insertions(+), 9 deletions(-)

diff --git a/bios_extract/src/ami.c b/bios_extract/src/ami.c
index 2d92e4b..e98d8ab 100644
--- a/bios_extract/src/ami.c
+++ b/bios_extract/src/ami.c
@@ -122,6 +122,25 @@ static char *AMI95ModuleNameGet(uint8_t ID)
 	return NULL;
 }
 
+static Bool
+IsAllZero(unsigned char *Buffer, int BufferSize)
+{
+	/* Optimized for x86, no idea about other architectures. */
+	register uintptr_t *ip = (uintptr_t *) Buffer;
+	register uintptr_t *iq = (uintptr_t *) (Buffer + BufferSize - (BufferSize % sizeof(uintptr_t)));
+	while (ip < iq) {
+		if (*ip++)
+			return FALSE;
+	}
+	register unsigned char *p = (unsigned char *) ip;
+	register unsigned char *q = Buffer + BufferSize;
+	while (p < q) {
+		if (*p++)
+			return FALSE;
+	}
+	return TRUE;
+}
+
 /*
  *
  */
@@ -218,9 +237,14 @@ NotCompressed:
 		if (Compressed) {
 			if (LH5Decode(BIOSImage + Offset + 8,
 				      ROMSize, Buffer, BufferSize) == -1) {
-				Compressed = FALSE;
+				Compressed = IsAllZero(Buffer, BufferSize);
 				munmap(Buffer, BufferSize);
-				unlink(filename);
+				if (Compressed) {
+					Compressed = FALSE;
+					unlink(filename);
+				} else {
+					strcpy(&filename[strlen(filename) - 3], "cmp");
+				}
 				goto NotCompressed;
 			}
 		} else
@@ -355,9 +379,14 @@ NotCompressed:
 		if (Compressed) {
 			if (LH5Decode(BIOSImage + ABCOffset + part->RealCS + 8,
 				      ROMSize, Buffer, BufferSize) == -1) {
-				Compressed = FALSE;
+				Compressed = IsAllZero(Buffer, BufferSize);
 				munmap(Buffer, BufferSize);
-				unlink(filename);
+				if (Compressed) {
+					Compressed = FALSE;
+					unlink(filename);
+				} else {
+					strcpy(&filename[strlen(filename) - 3], "cmp");
+				}
 				goto NotCompressed;
 			}
 		} else
@@ -579,9 +608,14 @@ NotCompressed:
 		if (Compressed) {
 			if (LH5Decode(BIOSImage + NewOffset,
 				      ROMSize, Buffer, BufferSize) == -1) {
-				Compressed = FALSE;
+				Compressed = IsAllZero(Buffer, BufferSize);
 				munmap(Buffer, BufferSize);
-				unlink(filename);
+				if (Compressed) {
+					Compressed = FALSE;
+					unlink(filename);
+				} else {
+					strcpy(&filename[strlen(filename) - 3], "cmp");
+				}
 				goto NotCompressed;
 			}
 		} else
diff --git a/biostools/analyzers.py b/biostools/analyzers.py
index ae0f709..caf828a 100644
--- a/biostools/analyzers.py
+++ b/biostools/analyzers.py
@@ -322,14 +322,15 @@ class AMIAnalyzer(Analyzer):
 		self._check_pattern = re.compile(b'''American Megatrends Inc|AMIBIOSC| Access Methods Inc\\.|AMI- ([0-9]{2}/[0-9]{2}/[0-9]{2}) (?:IBM is a TM of IBM|[\\x00-\\xFF]{2}   AMI-[^-]+-BIOS )''')
 		self._date_pattern = re.compile(b'''([0-9]{2}/[0-9]{2}/[0-9]{2})[^0-9]''')
 		self._uefi_csm_pattern = re.compile('''63-0100-000001-00101111-......-Chipset$''')
-		self._intel_86_pattern = re.compile('''(?:[0-9A-Z]{8})\.86(?:[0-9A-Z])\.(?:[0-9A-Z]{4})\.(?:[0-9A-Z]{3})\.(?:[0-9]{10})$''')
+		self._intel_86_pattern = re.compile('''(?:[0-9A-Z]{8})\\.86(?:[0-9A-Z])\\.(?:[0-9A-Z]{4})\\.(?:[0-9A-Z]{3})\\.(?:[0-9]{10})$''')
 		# The "All Rights Reserved" is important to not catch the same header on other files.
+		# "All<Rights Reserved" (Tatung TCS-9850 9600x9, corrupted during production?)
 		# AMIBIOS 6+ version corner cases:
 		# - Second digit not 0 (I forget which one had 000000)
 		# - Can be 4-digit instead of 6-digit (Biostar)
-		self._id_block_pattern = re.compile(b'''(?:AMIBIOS (?:(0[1-9][0-9]{2}[\\x00-\\xFF]{2})[\\x00-\\xFF]{2}|W ([0-9]{2}) ([0-9]{2})[\\x00-\\xFF])|0123AAAAMMMMIIII|\(AAMMIIBBIIOOSS\))([0-9]{2}/[0-9]{2}/[0-9]{2})\(C\)[0-9]{4} American Megatrends,? Inc(?:\.,? All Rights Reserved|/Hewlett-Packard Company)''')
+		self._id_block_pattern = re.compile(b'''(?:AMIBIOS (?:(0[1-9][0-9]{2}[\\x00-\\xFF]{2})[\\x00-\\xFF]{2}|W ([0-9]{2}) ([0-9]{2})[\\x00-\\xFF])|0123AAAAMMMMIIII|\\(AAMMIIBBIIOOSS\\))([0-9]{2}/[0-9]{2}/[0-9]{2})\\(C\\)[0-9]{4} American Megatrends,? Inc(?:\\.,?.All.Rights.Reserved|/Hewlett-Packard Company)''')
 		# Weird TGem identifier (TriGem 486-BIOS)
-		self._precolor_block_pattern = re.compile(b'''\\(C\\)(?:[0-9]{4}(?:AMI,404-263-8181|TGem-HCS,PSC,JGS)|( Access Methods Inc\.))''')
+		self._precolor_block_pattern = re.compile(b'''\\(C\\)(?:[0-9]{4}(?:AMI,404-263-8181|TGem-HCS,PSC,JGS)|( Access Methods Inc\\.))''')
 		# "Date:-" might not have a space after it (Intel AMI)
 		# "\xFF\xFF\xFF\xFFFLASH-" (Everex EISA 386-BIOS)
 		# Encoded "EVALUATION COPY" as a backup ("ami2939", possibly others without a date)